Graphics processor and acceleration method thereof

A graphics processor and memory technology, applied in processor architecture/configuration, image memory management, electrical digital data processing, etc., can solve problems such as time lengthening

Pending Publication Date: 2021-01-19
GLENFLY TECH CO LTD
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Traditional graphics processors need to perform two calculations to obtain the final matrix multiplication result, and the calculation and reasoning time becomes longer

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Graphics processor and acceleration method thereof
  • Graphics processor and acceleration method thereof
  • Graphics processor and acceleration method thereof

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0061]The present invention is described with reference to the drawings, wherein like reference numerals designate similar or identical elements throughout. The above-mentioned drawings are not drawn in actual scale, and are only provided to illustrate the present invention. Some aspects of the invention are described below as references to illustrate exemplary applications. This means that numerous specific details, relationships and methods are set forth in order to provide a complete understanding of the invention. However, one skilled in the art will recognize that the invention may be practiced without one or more of the specific details or otherwise. In other instances, well-known structures or operations are not listed in detail to avoid obscuring the invention. The invention is not limited by the illustrated acts or sequence of events, as some acts may occur in a different order or concurrently with other acts or events. Furthermore, not all recited acts or events n...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention provides a graphics processor and an acceleration method thereof. The graphics processor comprises a texel unit and an execution unit. The texel unit comprises a reading module. The execution unit comprises: an im2col module which is used for executing an im2col algorithm and expanding an original matrix according to the size of a convolution kernel to obtain an expanded matrix; an execution unit which multiplies the extended matrix by the convolution kernel to obtain a feature matrix; a reading module which calculates a characteristic coordinate of each element of the characteristic matrix according to the coordinate of the extended matrix, and obtains an original coordinate of each element of the original matrix according to the characteristic coordinate, the convolution kernel, the pace value and the filling value, wherein the reading module reads at least one of the storage blocks covered by the original coordinates of each element of the original matrix, and transmits data corresponding to the original coordinates in the at least one of the storage blocks to the im2col module.

Description

technical field [0001] The invention relates to a graphic processor, in particular to a graphic processor for improving im2col calculation efficiency and an acceleration method thereof. Background technique [0002] With the development of artificial intelligence, it has become an important direction to accelerate the neural network with a graphics processing unit (GPU). In order to hardware-accelerate neural network algorithms, various chip factories continue to improve the performance of computing modules, reduce data memory access, and design optimized algorithms, making the neural network reasoning performance faster and faster. General Matrix Multiplication (GEMM) is a widely used convolution acceleration method. Before executing GEMM, the input data needs to be prepared, and the process of preparing the input data is called image block rearrangement into matrix columns (im2col). [0003] When a traditional graphics processor executes GEMM, it must first read a matrix...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F17/15G06F17/16G06T1/20G06T1/60G06N3/063
CPCG06F17/15G06F17/16G06T1/20G06T1/60G06N3/063G06F12/0875G06F2212/454G06F2212/455G06F2212/1016G06F12/0897
Inventor 张炜顾德明
Owner GLENFLY TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap